Song DJ, et al. Allergy Asthma Immunol Res. 2018 Nov;10(6):591-613. https://doi.org/10.4168/aair.2018.10.6.591

Mycoplasma pneumoniae is a major causative pathogen of community-acquired pneumonia in children, and the treatment of choice is macrolides. There is an increasing trend in reports of refractory clinical responses despite macrolide treatment due to the emergence of macrolide-resistant M. pneumoniae. Early discrimination of macrolide-refractory M. pneumoniae pneumonia (MrMP) from macrolide-sensitive M. pneumoniae pneumonia (MSMP) is vital; however, testing for macrolide susceptibility at the time of admission is not feasible. This study aimed to identify the characteristics of MrMP in Korean children, in comparison with those of MSMP. In this multicenter study, board-certified pediatric pulmonologists at 22 tertiary hospitals reviewed the medical records from 2010 to 2015 of 5294 children who were hospitalized with M. pneumoniae pneumonia and administered macrolides as the initial treatment. One-way analysis of variance and the Kruskal-Wallis test were used to compare differences between groups. Of 5294 patients (mean age, 5.6 years) included in this analysis, 240 (4.5%), 925 (17.5%), and 4129 (78.0%) had MrMP, macrolide-less effective M. pneumoniae pneumonia, and MSMP, respectively. Compared with the MSMP group, the MrMP group had a longer fever duration, overall (13.0 days) and after macrolide use (8.0 days). A higher proportion of MrMP patients had respiratory distress, pleural effusion, and lobar pneumonia. The mean aspartate aminotransferase, alanine aminotransferase, lactate dehydrogenase, and C-reactive protein levels were the highest in the MrMP group, along with higher incidences of extrapulmonary manifestations and atelectasis (during and post infection). Pre-existing conditions were present in 17.4% (n = 725/4159) of patients, with asthma being the most common (n = 334/4811, 6.9%). This study verified that MrMP patients show more severe initial radiographic findings and clinical courses than MSMP patients. MrMP should be promptly managed by agents other than macrolides.

In this paper, we present an asthma risk prediction tool based on machine learning (ML). The entire tool is implemented on a smartphone as a mobile-health (m-health) application using the resources of Internet-of-Things (IoT). Peak Expiratory Flow Rates (PEFR) are commonly measured using external instruments such as peak flow meters and are well known asthama risk predictors. In this work, we find a correlation between the particulate matter (PM) found indoors and the outside weather with the PEFR. The PEFR results are classified into three categories such as ‘Green’ (Safe), ‘Yellow’ (Moderate Risk) and ‘Red’ (High Risk) conditions in comparison to the best peak flow value obtained by each individual. Convolutional neural network (CNN) architecture is used to map the relationship between the indoor PM and weather data to the PEFR values. The proposed method is compared with the state-of-the-art deep neural network (DNN) based techniques in terms of the root mean square and mean absolute error accuracy measures. These performance measures are better for the proposed method than other methods discussed in the literature. The entire setup is implemented on a smartphone as an app. An IoT system including a Raspberry Pi is used to collect the input data. This assistive tool can be a cost-effective tool for predicting the risk of asthma attacks.

Despite ample research on the association between indoor air pollution and allergic disease prevalence, public health and environmental policies still lack predictive evidence for developing a preventive guideline for patients or vulnerable populations mostly due to limitation of real-time big data and model predictability. Recent popularity of IoT and machine learning techniques could provide enabling technologies for collecting real-time big data and analyzing them for more accurate prediction of allergic disease risks for evidence-based intervention, but the effort is still in its infancy. This pilot study explored and evaluated the feasibility of a deep learning algorithm for predicting asthma risk. It is based on peak expiratory flow rates (PEFR) of 14 pediatric asthma patients visiting the Korea University Medical Center and indoor particulate matter PM10 and PM2.5 concentration data collected at their residence every 10 minutes using a PM monitoring device with a low-cost sensor between September 1, 2017 and August 31, 2018. We interpolated the PEFR results collected twice a day for each patient throughout the day so that it can be matched to the PM and other weather data. The PEFR results were classified into three categories such as ‘Green’ (normal), ‘Yellow’ (mild to moderate exacerbation) and ‘Red’ (severe exacerbation) with reference to their best peak flow value. Long Short-Term Memory (LSTM) model was trained using the first 10 months of the linked data and predicted asthma risk categories for the next 2 months during the study period. LSTM model is found to predict the asthma risk categories better than multinomial logistic (MNL) regression as it incorporates the cumulative effects of PM concentrations over time. Upon successful modifications of the algorithm based on a larger sample, this approach could potentially play a groundbreaking role for the scientific data-driven medical decision making.

Song DJ, et al. Allergy Asthma Immunol Res. 2017 Jul;9(4):360-367. https://doi.org/10.4168/aair.2017.9.4.360

Seo YH, et al. Korean J Pediatr. 2014 Apr;57(4):186-192. http://dx.doi.org/10.3345/kjp.2014.57.4.186